Research Of Infrared Image Detection Method On Mine Conveyor Belt Longitudinal Tear

Mine conveyor belt is the most economical means with high tensile strength, little elongation, strong transport capacity and high carrying power in mine transportation. However, because of the long time and high load work status and happening of accident, longitudinal tear accident often happens, which may cause incalculable economic loss to the company and the serious casualties. Therefore, the longitudinal tear fault detection is always being the important research subject of coal mine safety. The existing detection methods at home and abroad can be divided into two categories according to the detection methods:contact detection method and non-contact detection method. But the above methods have the defect of poor real-time performance and high motion error rate. And testing instrument has also not been considered that is unable to overcome the influence of coal mine complex environment, so the testing result is not accurate. This article elaborates the present test methods of mine conveyor belt longitudinal tearing at home and abroad, studying the structure of mine conveyor and the reason of longitudinal tear, researching the principle of infrared image detection method on mine conveyor belt longitudinal tear, Analysis the existing method of infrared image segmentation, which Includes: histogram twin peaks method, iteration method, the maximum entropy method and the method based on BP neural network,through the theoretical analysis and experiments show its defects, proposing infrared image segmentation based on SVM used in conveyer belt longitudinal tear test, comparative studying the infrared image segmentation result between, SVM and BP neural network, and researching the infrared image detection method on mine conveyor belt longitudinal tear based on above principle. A detailed description on the hardware selection of the test method is made. In the software, the article takes virtual instrument as software architecture. Three ways of infrared camera driven by LabVIEW is described. Infrared image segmentation algoritm program based on SVM is introduced in details, which is used by Matlab. The ways of LabVIEW and Matlab being mixed to programme are introduced and researched. Then the most suitable programming method is chosen.The whole program of software is introduced, including:main VI, project and realtime image save Ⅵ. Infrared camera is controlled by LabVIEW to collect infrared image and these images are feedback to Matlab to be segmented by SVM. The processed results are inputed into LabVIEW and the corresponding processing will be done by LabVIEW.Finally the simulation tests about above methods, including real-time anti-interference and accuracy analysis, are made. Through the simulation experiments, this method has been proved with ability to overcome the low accuracy, good real-time performance caused by the environmental impact effectively. And it has strong identification and portability ability, which can satisfy the requirement of the conveyor belt longitudinal tearing detection. This article provides a new theory and a practical method on conveyer belt longitudinal tear test with infrared image.